Automated Writing Evaluation and Generative AI in Education: An Integrative Review of Efficacy, Pedagogy, and Technology

Talebi, Mehran; Taghipour, Masoud

doi:10.22034/quipls.2026.2091335.1037

Automated Writing Evaluation and Generative AI in Education: An Integrative Review of Efficacy, Pedagogy, and Technology

Articles in Press

Document Type : Review Article

Authors

Mehran Talebi ¹

Masoud Taghipour ²

¹ English Language and Literature, Persian Literature and foreign languages, Allameh Tabataba'i University, Tehran, Iran.

² English Language and Literature, Persian Literature and foreign languages, Allameh Tabataba'i University, Tehran, Iran.

10.22034/quipls.2026.2091335.1037

Abstract

Automated Writing Evaluation (AWE) has moved from rule-based correction and automated scoring to generative tools based on large language models. This integrative literature review synthesises 35 empirical studies, reviews, meta-analyses, and bibliometric analyses published between 2002 and 2025. The results indicate four major patterns: AWE can improve accuracy and practice volume when learners engage in repeated drafting and revision; implementation depends strongly on teacher orchestration, student perceptions, and classroom ecology; technological development is moving from micro-level correction toward genre-aware, process-sensitive, and LLM-supported feedback; and Generative AI creates a new distinction between performance with a tool and learning that transfers after the tool is removed. The discussion interprets these findings as evidence that AWE should be treated as a human-in-the-loop pedagogy rather than an autonomous replacement for teachers. Future research should focus on active ingredients, feedback literacy, explainable AI, validity, and cognitive residue in order to determine not only whether writing improves during tool use but also whether durable writing ability develops.

Keywords

Automated Writing Evaluation

Efficacy

Generative AI

Writing Pedagogy

Teacher Agency

Subjects